2.3.2.5 Algorithmically neentloni
Ukuziphatha kwi wafumana yedatha ongowemvelo, iqhutywa iinjongo zobunjineli kweenkqubo.
Nangona imithombo yedatha wafunyanwa abaninzi non-yovukelo kuba abantu azisilindelanga data zabo zibhalisiwe (Section 2.3.1.3), abaphandi kufuneka bacinge ukuziphatha kwezi nkqubo intanethi ukuba "zisenzeka ngokwendalo" okanye "nyulu." Enyanisweni, le iinkqubo yesuntswana ukuba ukuziphatha irekhodi ziyangqamana ngobunjineli ukubaphembelela imikhwa ezithile ezifana ngokunqakraza iintengiso okanye ngeposi umxholo. Iindlela ukuba iinjongo abayili inkqubo onokwazisa iipateni zibe data kuthiwa neyokuqungaqunga algorithmic. neyokuqungaqunga Algorithmic awaziwa noko ntle, kodwa yinkxalabo enkulu phakathi izazinzulu data ngenyameko. Yaye, ngokungafaniyo ezinye iingxaki kunye imizila digital, neyokuqungaqunga algorithmic becala ezingabonakaliyo.
Umzekelo olulula neyokuqungaqunga algorithmic kukuba on Facebook kukho inani anomalously eliphezulu abasebenzisi kunye nama-20 abahlobo (Ugander et al. 2011) . Izazinzulu nokuhlalutya nale data ngaphandle ingqondo kukaFacebook isebenza njani doubtlessly ukujikelezisa amabali amaninzi malunga nendlela 20 ezinye uhlobo inani imilingo zentlalo. Nangona kunjalo, Ugander noogxa bakhe baziqonda enkulu inkqubo uvelise data, baza baqonda ukuba Facebook wakhuthaza abantu ngendibaniso abambalwa kwiFacebook ukwenza abahlobo de afika abahlobo-20. Nangona Ugander noogxa musani ukuthi oku phepha, lo mgaqo kwakuyeyona wadalwa nguFacebook ukwenzela ukukhuthaza abasebenzisi abatsha ukuba bazibandakanye. Ngaphandle besazi ubukho kwalo mgaqo-nkqubo, kunjalo, kulula ukuzoba kwisigqibo engalunganga evela data. Ngamanye amazwi, inani Akumangalisi eliphezulu labantu nabahlobo 20 isixelela okungakumbi Facebook ngaphezu ukuziphatha kwabantu.
Okuninzi iyingozi ngaphezu kulo mzekelo edlulileyo apho neyokuqungaqunga algorithmic livelise japan quirky ukuba abaphandi ngenyameko ukuze uphando, kukho kuguqulelo nokuba trickier neyokuqungaqunga algorithmic ukuba kwenzeka xa abayili lweenkqubo online banolwazi iingcamango zentlalo babhake ezi ngcamango zibe ngokokusebenza kweenkqubo zabo. Ntle kubiza le performativity: xa neengcamango ukutshintsha ihlabathi ngendlela wokuba bakuzisele ihlabathi ezininzi ngokuhambelana le ngcamango. Xa iimeko neyokuqungaqunga adlalwayo algorithmic, uhlobo lidane lwedata kusenokwenzeka ezingabonakaliyo.
Omnye umzekelo iphatheni eyenziwe yi performativity na transitivity kubuxhakaxhaka zentlalo online. Kwi 1970 and 1980, ngokuphindaphindiweyo abaphandi bafumanise ukuba xa abahlobo kunye Alice kwaye wena nabahlobo linguBob, ke uMonde Alice kungenzeka ukuba babe zizihlobo ngaphandle abantu ababini jikelele esikhethiweyo. Ke, le ndlela inye kakhulu eyafunyanwa grafu kwezentlalo Facebook (Ugander et al. 2011) . Ngoko, umntu usenokugqiba ukuba izizekelo ubuhlobo kwiFacebook ekunabiseleni iipateni bahlobo oflayini, ubuncinane ngokwemiqathango transitivity. Nangona kunjalo, inqontsonqa transitivity kwi-Facebook grafu zentlalo engagqibekanga iqhutywa neyokuqungaqunga algorithmic. Oko kukuthi, izazinzulu data kula Facebook wayesazi lokufumanisa zethiyori uphando malunga transitivity kugutywa ngayo Facebook isebenza njani. Facebook kukho "Abantu Ngamana Yazi" osemqoka uuba abahlobo abatsha, kwaye enye indlela facebook isigqibo ngubani ukucebisa kuni transitivity. Oko kukuthi, Facebook kungenzeka ukuba bacebise ukuba babe ngabahlobo nabo bahlobo nabahlobo bakho. lo enjenje osemqoka kunesiphumo yokwandisa transitivity kwi-Facebook grafu zentlalo; ngamanye amazwi, le ngcamango transitivity kuzisa ihlabathi ngokuhambelana kunye neengqikelelo kwengcamango (Healy 2015) . Ngoko, xa imithombo yedatha enkulu ibonakala lokuyivelisa iingqikelelo ithiyori zentlalo, kufuneka siqiniseke ukuba le ngcamango ngokwayo akazange kugutywa kakuhle inkqubo.
Kunokuba ucinga imithombo yedatha enkulu njengoko ibona abantu ngendlela yemvelo, isikweko babekulangazelela sigcina abantu lokungcakaza. Iikhasino ziyangqamana ngobunjineli bume yenzelwe ukubaphembelela imikhwa ethile, yaye abaphandi soze silindele ukuba ukuziphatha lokungcakaza kungabonelela window unfettered kwihambo yabantu. Kakade ke, siya kukwazi ukufunda into ethile malunga nokuziphatha ukufunda abantu ngabantu kwiikhasino-enyanisweni ekhasino kunokuba kuba elihle ukufunda ulwalamano phakathi ukuselwa kotywala kunye nomngcipheko ukhetho-kodwa xa si tyeshela ukuba idatha kwa kudalwa kwikhasino sibe nokuba uzobe ezinye izigqibo ezimbi.
Ngelishwa, ukumelana neyokuqungaqunga algorithmic kunzima ngakumbi kuba iimpawu ezininzi iinkqubo zakwi apho, zilambatha kumaxwebhu, kwaye ezihlala ziguquka. Ngokomzekelo, njengoko ndiza ukucacisa kamva kwesi sahluko, neyokuqungaqunga algorithmic yaba ingcaciso elinye kunokwenzeka ngenxa ngcembe ikhefu-phantsi Google Flu Iintsingiselo (Section 2.4.2), kodwa eli bango Kwakunzima ukuhlola kuba intsebenziso zangaphakathi ze zophendlo Google ngayo algorithm na apho. Uhlobo sokuguquguquka neyokuqungaqunga algorithmic yenye uhlobo bokuphaphatheka kwinkqubo. neyokuqungaqunga Algorithmic kuthetha ukuba kufuneka balumke malunga naluphi na ibango ukuziphatha kwabantu olubangelwa inkqubo yedijithali enye, nokuba likhulu kangakanani.
